At the current level of autonomous vehicle abilities, I agree with you, in a broad sense. Vehicles will need to still be able to differentiate between shapes, even during bad weather. Weather like blizzards, sudden downpours, heavy fog, dust storms, and the like. You still have to be able to see to safely pull off of the road.

Until we can guarantee with 100% certainty that they can truly drive without aid, I completely agree that these vehicles would not be safe on their own. Weather is very well known for being unpredictable at times. Life in general is also known for being unpredictable at times.

What happens if the sensors are unknowingly damaged? What happens if someone is wearing a costume that makes them look like a giant cereal box instead of human-shaped? What happens if there’s a software glitch at a bad time? What protections are there to guarantee that it doesn’t happen? Are those protections temporary? How often should they be reviewed?

It should be OK to acknowledge that we aren’t quite there yet. Yes, it seems cool and all, but it’s silly to risk lives over impatience. If it will happen, it will happen. Forcing it to happen sooner than it should could very well lead to it being banned altogether, especially if enough people die or get injured as a result.

IMO, anyone who causes serious crashes from using these things in “fully autonomous” mode should be charged as if the vehicle wasn’t autonomous. As if the accident was caused by sleeping behind the wheel or texting while driving. The company should be charged similarly in that scenario, as their programming and marketing would also play a part in the crash.

Hey, if they’re truly safe, none of these charges would actually happen. If there isn’t an “oops” death in the first place, there won’t be an “oops” death to investigate.